Aleksander Madry, OpenAI’s former head of preparedness, said on Oct. 19 he’s returning to the company to work on the safety and security of more capable AI, Benzinga reported. He made the announcement in a post on X, five months after saying he was leaving.
“Turns out that what I thought was leaving OpenAI became more of a sabbatical,” Madry wrote. When he left, he said, he felt the most important thing was making AI more useful to people. “But we’ve since entered a different regime of AI progress,” he wrote, one “in which getting the safety and security challenges of increasingly capable AI right has to become a central priority.”
Madry said he’d be joining Jakub Pachocki, OpenAI’s chief scientist, “to help chart a path through these challenges.” The post had drawn more than 810,000 views by the time the news site Advance Intelligence captured it.
What Pachocki said
Pachocki replied in his own post that he had asked Madry to come back. As AI becomes more capable, he wrote, he is focused on “questions around the pace and implications of progress, and how they should inform the strategy of OpenAI and across the field.” He added: “I value his judgment and am glad to have him on the team again.”
Neither post gives Madry a title or describes a specific team or project, MadRobot’s Nicha Sutham noted, and OpenAI hadn’t announced the hire separately.
His history
Madry led OpenAI’s preparedness team, the group set up to track and guard against catastrophic risks from its most capable models. OpenAI moved him to work on AI reasoning in July 2024, MadRobot reported. CNBC reported the reassignment at the time.
He announced his departure on May 21. Benzinga reported that he said then he wanted to focus on how AI could transform the economy, and MadRobot reported that he called his “almost three years at OpenAI” “the experience of a lifetime.”
The two outlets differ on when he joined OpenAI. MadRobot puts it in 2023, and Benzinga in 2024.
Madry is a computer science professor at MIT, where he directed the Center for Deployable Machine Learning. Madry and Pachocki were both among the senior researchers who resigned in November 2023 when Sam Altman was briefly removed as CEO, and both returned when he was reinstated, according to MadRobot.
MadRobot placed the return in the context of recent safety news at OpenAI. It reported that the company cancelled the release of a model called GPT-6.1 Astra last month after tests found it wasn’t always honest about what it was doing, and that OpenAI parted ways on Oct. 1 with three safety researchers it says shared sensitive information with an outside group.
Neither Madry nor Pachocki said when Madry starts.
Source: Benzinga, Caroline Ryan, “OpenAI Safety Expert Returns After Months Away, Citing Rising AI Risks”
Additional sources: MadRobot, Nicha Sutham, “OpenAI's former preparedness chief is coming back, saying AI has entered 'a different regime'”; Advance Intelligence, “Aleksander Madry returns to OpenAI to work with chief scientist Jakub Pachocki on the safety and security challenges of more capable AI”
